1: \begin{abstract}

2: This paper considers several linear beamformer design paradigms

3: for multiuser time-invariant multiple-input multiple-output

4: interference channels. Notably,   interference alignment and

5: sum-rate based algorithms such as the maximum

6: signal-to-interference-plus noise (max-SINR) algorithm are

7: considered.  Optimal linear beamforming under interference

8: alignment consists of two layers; an inner precoder and decoder

9: (or receive filter) accomplish interference alignment to eliminate

10: inter-user interference, and an outer precoder and decoder

11: diagonalize the effective single-user channel resulting from the

12: interference alignment by the inner precoder and decoder.  The

13: relationship between this two-layer beamforming  and the max-SINR

14: algorithm is established at high signal-to-noise ratio. Also, the

15: optimality of the max-SINR algorithm within the class of linear

16: beamforming algorithms, and its local convergence with exponential

17: rate, are established at high signal-to-noise ratio.

18: \end{abstract}